302 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3=================4Boot-time tracing5=================6 7:Author: Masami Hiramatsu <mhiramat@kernel.org>8 9Overview10========11 12Boot-time tracing allows users to trace boot-time process including13device initialization with full features of ftrace including per-event14filter and actions, histograms, kprobe-events and synthetic-events,15and trace instances.16Since kernel command line is not enough to control these complex features,17this uses bootconfig file to describe tracing feature programming.18 19Options in the Boot Config20==========================21 22Here is the list of available options list for boot time tracing in23boot config file [1]_. All options are under "ftrace." or "kernel."24prefix. See kernel parameters for the options which starts25with "kernel." prefix [2]_.26 27.. [1] See :ref:`Documentation/admin-guide/bootconfig.rst <bootconfig>`28.. [2] See :ref:`Documentation/admin-guide/kernel-parameters.rst <kernelparameters>`29 30Ftrace Global Options31---------------------32 33Ftrace global options have "kernel." prefix in boot config, which means34these options are passed as a part of kernel legacy command line.35 36kernel.tp_printk37 Output trace-event data on printk buffer too.38 39kernel.dump_on_oops [= MODE]40 Dump ftrace on Oops. If MODE = 1 or omitted, dump trace buffer41 on all CPUs. If MODE = 2, dump a buffer on a CPU which kicks Oops.42 43kernel.traceoff_on_warning44 Stop tracing if WARN_ON() occurs.45 46kernel.fgraph_max_depth = MAX_DEPTH47 Set MAX_DEPTH to maximum depth of fgraph tracer.48 49kernel.fgraph_filters = FILTER[, FILTER2...]50 Add fgraph tracing function filters.51 52kernel.fgraph_notraces = FILTER[, FILTER2...]53 Add fgraph non-tracing function filters.54 55 56Ftrace Per-instance Options57---------------------------58 59These options can be used for each instance including global ftrace node.60 61ftrace.[instance.INSTANCE.]options = OPT1[, OPT2[...]]62 Enable given ftrace options.63 64ftrace.[instance.INSTANCE.]tracing_on = 0|165 Enable/Disable tracing on this instance when starting boot-time tracing.66 (you can enable it by the "traceon" event trigger action)67 68ftrace.[instance.INSTANCE.]trace_clock = CLOCK69 Set given CLOCK to ftrace's trace_clock.70 71ftrace.[instance.INSTANCE.]buffer_size = SIZE72 Configure ftrace buffer size to SIZE. You can use "KB" or "MB"73 for that SIZE.74 75ftrace.[instance.INSTANCE.]alloc_snapshot76 Allocate snapshot buffer.77 78ftrace.[instance.INSTANCE.]cpumask = CPUMASK79 Set CPUMASK as trace cpu-mask.80 81ftrace.[instance.INSTANCE.]events = EVENT[, EVENT2[...]]82 Enable given events on boot. You can use a wild card in EVENT.83 84ftrace.[instance.INSTANCE.]tracer = TRACER85 Set TRACER to current tracer on boot. (e.g. function)86 87ftrace.[instance.INSTANCE.]ftrace.filters88 This will take an array of tracing function filter rules.89 90ftrace.[instance.INSTANCE.]ftrace.notraces91 This will take an array of NON-tracing function filter rules.92 93 94Ftrace Per-Event Options95------------------------96 97These options are setting per-event options.98 99ftrace.[instance.INSTANCE.]event.GROUP.EVENT.enable100 Enable GROUP:EVENT tracing.101 102ftrace.[instance.INSTANCE.]event.GROUP.enable103 Enable all event tracing within GROUP.104 105ftrace.[instance.INSTANCE.]event.enable106 Enable all event tracing.107 108ftrace.[instance.INSTANCE.]event.GROUP.EVENT.filter = FILTER109 Set FILTER rule to the GROUP:EVENT.110 111ftrace.[instance.INSTANCE.]event.GROUP.EVENT.actions = ACTION[, ACTION2[...]]112 Set ACTIONs to the GROUP:EVENT.113 114ftrace.[instance.INSTANCE.]event.kprobes.EVENT.probes = PROBE[, PROBE2[...]]115 Defines new kprobe event based on PROBEs. It is able to define116 multiple probes on one event, but those must have same type of117 arguments. This option is available only for the event which118 group name is "kprobes".119 120ftrace.[instance.INSTANCE.]event.synthetic.EVENT.fields = FIELD[, FIELD2[...]]121 Defines new synthetic event with FIELDs. Each field should be122 "type varname".123 124Note that kprobe and synthetic event definitions can be written under125instance node, but those are also visible from other instances. So please126take care for event name conflict.127 128Ftrace Histogram Options129------------------------130 131Since it is too long to write a histogram action as a string for per-event132action option, there are tree-style options under per-event 'hist' subkey133for the histogram actions. For the detail of the each parameter,134please read the event histogram document (Documentation/trace/histogram.rst)135 136ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]keys = KEY1[, KEY2[...]]137 Set histogram key parameters. (Mandatory)138 The 'N' is a digit string for the multiple histogram. You can omit it139 if there is one histogram on the event.140 141ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]values = VAL1[, VAL2[...]]142 Set histogram value parameters.143 144ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]sort = SORT1[, SORT2[...]]145 Set histogram sort parameter options.146 147ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]size = NR_ENTRIES148 Set histogram size (number of entries).149 150ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]name = NAME151 Set histogram name.152 153ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]var.VARIABLE = EXPR154 Define a new VARIABLE by EXPR expression.155 156ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]<pause|continue|clear>157 Set histogram control parameter. You can set one of them.158 159ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]onmatch.[M.]event = GROUP.EVENT160 Set histogram 'onmatch' handler matching event parameter.161 The 'M' is a digit string for the multiple 'onmatch' handler. You can omit it162 if there is one 'onmatch' handler on this histogram.163 164ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]onmatch.[M.]trace = EVENT[, ARG1[...]]165 Set histogram 'trace' action for 'onmatch'.166 EVENT must be a synthetic event name, and ARG1... are parameters167 for that event. Mandatory if 'onmatch.event' option is set.168 169ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]onmax.[M.]var = VAR170 Set histogram 'onmax' handler variable parameter.171 172ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]onchange.[M.]var = VAR173 Set histogram 'onchange' handler variable parameter.174 175ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]<onmax|onchange>.[M.]save = ARG1[, ARG2[...]]176 Set histogram 'save' action parameters for 'onmax' or 'onchange' handler.177 This option or below 'snapshot' option is mandatory if 'onmax.var' or178 'onchange.var' option is set.179 180ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.[N.]<onmax|onchange>.[M.]snapshot181 Set histogram 'snapshot' action for 'onmax' or 'onchange' handler.182 This option or above 'save' option is mandatory if 'onmax.var' or183 'onchange.var' option is set.184 185ftrace.[instance.INSTANCE.]event.GROUP.EVENT.hist.filter = FILTER_EXPR186 Set histogram filter expression. You don't need 'if' in the FILTER_EXPR.187 188Note that this 'hist' option can conflict with the per-event 'actions'189option if the 'actions' option has a histogram action.190 191 192When to Start193=============194 195All boot-time tracing options starting with ``ftrace`` will be enabled at the196end of core_initcall. This means you can trace the events from postcore_initcall.197Most of the subsystems and architecture dependent drivers will be initialized198after that (arch_initcall or subsys_initcall). Thus, you can trace those with199boot-time tracing.200If you want to trace events before core_initcall, you can use the options201starting with ``kernel``. Some of them will be enabled eariler than the initcall202processing (for example,. ``kernel.ftrace=function`` and ``kernel.trace_event``203will start before the initcall.)204 205 206Examples207========208 209For example, to add filter and actions for each event, define kprobe210events, and synthetic events with histogram, write a boot config like211below::212 213 ftrace.event {214 task.task_newtask {215 filter = "pid < 128"216 enable217 }218 kprobes.vfs_read {219 probes = "vfs_read $arg1 $arg2"220 filter = "common_pid < 200"221 enable222 }223 synthetic.initcall_latency {224 fields = "unsigned long func", "u64 lat"225 hist {226 keys = func.sym, lat227 values = lat228 sort = lat229 }230 }231 initcall.initcall_start.hist {232 keys = func233 var.ts0 = common_timestamp.usecs234 }235 initcall.initcall_finish.hist {236 keys = func237 var.lat = common_timestamp.usecs - $ts0238 onmatch {239 event = initcall.initcall_start240 trace = initcall_latency, func, $lat241 }242 }243 }244 245Also, boot-time tracing supports "instance" node, which allows us to run246several tracers for different purpose at once. For example, one tracer247is for tracing functions starting with "user\_", and others tracing248"kernel\_" functions, you can write boot config as below::249 250 ftrace.instance {251 foo {252 tracer = "function"253 ftrace.filters = "user_*"254 }255 bar {256 tracer = "function"257 ftrace.filters = "kernel_*"258 }259 }260 261The instance node also accepts event nodes so that each instance262can customize its event tracing.263 264With the trigger action and kprobes, you can trace function-graph while265a function is called. For example, this will trace all function calls in266the pci_proc_init()::267 268 ftrace {269 tracing_on = 0270 tracer = function_graph271 event.kprobes {272 start_event {273 probes = "pci_proc_init"274 actions = "traceon"275 }276 end_event {277 probes = "pci_proc_init%return"278 actions = "traceoff"279 }280 }281 }282 283 284This boot-time tracing also supports ftrace kernel parameters via boot285config.286For example, following kernel parameters::287 288 trace_options=sym-addr trace_event=initcall:* tp_printk trace_buf_size=1M ftrace=function ftrace_filter="vfs*"289 290This can be written in boot config like below::291 292 kernel {293 trace_options = sym-addr294 trace_event = "initcall:*"295 tp_printk296 trace_buf_size = 1M297 ftrace = function298 ftrace_filter = "vfs*"299 }300 301Note that parameters start with "kernel" prefix instead of "ftrace".302